How to Use Claude MCP to Schedule Social Media Posts

To schedule social media posts with Claude MCP, add your scheduler's remote MCP server to Claude as a custom connector, sign in with OAuth, and ask Claude to draft, check, and schedule posts in the chat. Custom connectors work on every Claude plan, including Free with one connector, and Claude Code can add the same server from the terminal.

Claude does not publish to Instagram or LinkedIn by itself. It calls tools that your scheduling app exposes over the Model Context Protocol, and the app handles the network connections, the queue, and the publishing.

This guide explains what Claude MCP is, how each Claude app connects to MCP servers as of October 2026, and how to set up a working flow with PostQuest, the scheduler I build. It also covers the prompts that work well, how approval works, what to do when the connection fails, and where the setup falls short.

I checked every step in this guide against Anthropic's help center, the Claude Code documentation, and the PostQuest code in October 2026. The Claude side works the same with any scheduler that runs a remote MCP server. Only the server URL and the tool names differ.

What is Claude MCP?

Claude MCP is Claude's support for the Model Context Protocol, an open standard that lets Claude call tools and read data from outside apps. The MCP documentation describes MCP as "an open-source standard for connecting AI applications to external systems" and compares it to a USB-C port for AI apps. Anthropic open-sourced the protocol on November 25, 2024.

An MCP setup has two sides. The MCP server belongs to the app, such as a scheduler, and publishes a list of tools with names, descriptions, and inputs. The MCP client is Claude, which reads that list and decides which tools to call while it works on your request.

For social media, this means Claude can see your connected accounts, write a draft into your scheduler, check it against each network's rules, and put it on the calendar. You stay in the conversation, and the scheduler does the publishing. Claude never receives your Instagram or LinkedIn password. It only gets the access you grant to the scheduler's MCP server.

How does Claude connect to MCP servers in 2026?

Claude connects to remote MCP servers through custom connectors in claude.ai, Claude Desktop, Cowork, and the mobile apps, and through the claude mcp add command in Claude Code. Both paths use the same remote server URL. They differ in who can add the server and where the connection comes from.

According to Anthropic's custom connectors help article (updated October 1, 2026), custom connectors are available on Free, Pro, Max, Team, and Enterprise plans, and Free users are limited to one custom connector. Claude reaches the server from Anthropic's cloud, not from your computer, so the server must be reachable on the public internet. A hosted scheduler meets that requirement.

Where you use Claude	How you add a server	Plan rules (October 2026)
claude.ai and the web	Customize → Connectors → + Add → Add custom connector	Free (one connector), Pro, Max
Claude Desktop and Cowork	Same connector settings, synced through your Claude account	Same as claude.ai
Team and Enterprise	An owner adds it in Organization settings → Connectors	Members then click Connect in Customize → Connectors
Mobile apps	Connectors on your Claude account work here too	Same account rules
Claude Code	claude mcp add --transport http <name> <url>, or reuse your claude.ai connectors	Any account that can run Claude Code

Claude Desktop also supports local MCP servers configured in claude_desktop_config.json. That is a separate mechanism for programs that run on your own machine. You do not need it for a hosted scheduler.

What you need before you start

Check four things before you open Claude's settings:

• A Claude account. Any plan works for one connector. On Team or Enterprise, ask an owner to add the connector first.
• A PostQuest account with a finished onboarding. OAuth sign-in only works for an account that has verified its email and completed setup. The seven-day trial includes MCP access.
• At least one connected social account in the workspace you plan to use. Claude can only schedule to accounts that are already connected in PostQuest.
• The MCP endpoint: https://mcp.post-quest.com/mcp. It is also shown in PostQuest under Settings → API & MCP and on the PostQuest MCP page for Claude.

How do you connect PostQuest to Claude?

To connect PostQuest to Claude, add https://mcp.post-quest.com/mcp as a custom connector, sign in with OAuth, and authorize one workspace. In claude.ai or Claude Desktop, open Customize → Connectors, click + Add, choose Add custom connector, name it PostQuest, and paste the URL. Under Authentication, choose Sign in now, set the OAuth client to Register automatically, and leave Request headers empty. A PostQuest window opens where you sign in, pick the workspace Claude may use, and click Authorize. Then switch PostQuest on for each chat from the + menu. In Claude Code, run claude mcp add --transport http postquest https://mcp.post-quest.com/mcp, then finish the browser sign-in from /mcp. Claude shows each tool call it makes in the chat. The setup takes a few minutes and needs no API key. Claude can then draft and validate posts, schedules them only with your approval, and never receives your passwords.

Connect in claude.ai or Claude Desktop

1. Open Customize → Connectors.
2. Click + Add, then Add custom connector.
3. Enter PostQuest as the name and paste https://mcp.post-quest.com/mcp as the remote MCP server URL. Click Continue.
4. Review the authentication settings Claude detects and click Continue. Then set three fields and click Add:
   • Authentication: choose Sign in now.
   • OAuth client: choose Register automatically. PostQuest supports automatic client registration.
   • Request headers: leave empty. PostQuest uses OAuth, not a fixed key.
5. A browser window opens on PostQuest. Sign in, choose the workspace Claude may access, and click Authorize. If it does not open, click Connect next to PostQuest in Customize → Connectors.
6. In a new chat, click the + button at the lower left, open Connectors, and switch PostQuest on for that conversation.

The consent screen states what Claude can do: read accounts and posts, manage media and drafts, and schedule posts. Immediate publishing stays off unless you turn it on in that workspace's settings.

Connect on a Team or Enterprise plan

An owner or primary owner opens Organization settings → Connectors, clicks Add, hovers over Custom, selects Web, and enters the same name and URL. On Enterprise, people with a custom role that includes Manage access to Libraries can do this too, according to Anthropic's connector article. Each member then opens Customize → Connectors, finds PostQuest, and clicks Connect to sign in with their own PostQuest account. Each person authorizes their own workspace, so Claude sees only what that person can see.

Connect in Claude Code

If you log in to Claude Code with your claude.ai account, the connectors you added on claude.ai already appear there. Run /mcp to check. To add PostQuest directly instead, run:

claude mcp add --transport http postquest https://mcp.post-quest.com/mcp

Then run /mcp inside Claude Code, choose the server, and finish the browser sign-in. The Claude Code MCP documentation also describes claude mcp login <name> for signing in from the shell and the --scope user flag for making the server available in all your projects.

For scripts and headless runs, PostQuest also offers a direct endpoint at https://mcp.post-quest.com/mcp/direct that accepts an API key in an Authorization: Bearer header. Create the key under Settings → API & MCP. Keys are bound to one workspace and expire after 90 days.

What can Claude do in PostQuest?

Once connected, Claude sees a fixed set of PostQuest tools. It cannot do anything outside that list. The groups below match the tool catalog on the PostQuest MCP hub.

Job	Tools	What they do
Discover your workspace	list_social_accounts, get_platform_requirements	Read connected accounts, workspace time zone, content pillars, and each network's caption and media rules
Draft and refine	create_post, update_post, set_platform_captions, validate_post	Create a draft, edit it, write a different caption per network, and dry-run publish-time checks
Schedule and publish	schedule_post, publish_post	Put an approved post on the calendar, or publish it now if your workspace allows that
Review and manage	list_posts, get_post, delete_post	See what is queued, inspect one post, and delete a draft or scheduled post
Manage media	upload_media, open_media_uploader, list_media	Import an image or video from a public URL, or open an upload widget for files on your device
Measure results	get_analytics_summary, list_post_analytics, get_post_analytics, sync_analytics	Read per-post metrics synced from the networks

Two details save time. First, validate_post is read-only. It runs the real publishing checks inside a transaction that is rolled back, so it never saves a post or uses your quota. Second, a file you attach to a Claude chat cannot be passed to PostQuest directly. Claude opens the media uploader instead, and you pick the file from your device. In a terminal client such as Claude Code, which cannot show the widget, Claude gives you an upload link to open in the browser.

Which prompts work, and what does Claude do with each?

The best prompts name the accounts, the goal, and the boundary. "Keep everything as drafts" or "ask before scheduling" makes the approval step explicit. The table shows the tool calls you can expect. Claude chooses the exact sequence, and it shows each call in the conversation.

Prompt	Tools Claude calls	Result
"Which accounts are connected, and does any of them need reconnecting?"	list_social_accounts	A list of accounts with any that need a reconnect flagged, plus your workspace time zone
"Propose five posts for next week from my content pillars. Save them as drafts."	list_social_accounts, get_platform_requirements, create_post (five times)	Five drafts in PostQuest. Nothing is scheduled
"Adapt this announcement for LinkedIn, Threads, and Bluesky without changing the main claim."	create_post, set_platform_captions, validate_post	One post with three tailored captions and a list of any blocking issues
"Schedule the LinkedIn draft for Tuesday at 9:00 a.m. New York time."	get_post, validate_post, then schedule_post after you approve	A scheduled post with an ISO-8601 time such as 2026-10-13T09:00:00-04:00
"Which of last month's posts did best on Instagram, and what do they have in common?"	get_analytics_summary, list_post_analytics	A ranked list with Claude's read on format and topic patterns

Claude writes the captions with its own model. PostQuest stores, checks, and publishes them. That split matters when you want Claude to repurpose one idea across networks. The method in turn one content idea into five posts works well as a prompt: give Claude the source idea and ask for one draft per network.

A worked example: one launch post for LinkedIn and X

Here is what a typical session looks like.

1. You ask: "Draft a launch post for LinkedIn and X about our new export feature. Keep it as a draft."
2. Claude calls list_social_accounts to find the account IDs, then get_platform_requirements for both networks. It learns, for example, that a standard X account allows 280 characters.
3. Claude calls create_post with both accounts and then set_platform_captions to give LinkedIn a longer version and X a short one.
4. Claude calls validate_post. If the X caption is too long or an image has the wrong aspect ratio, the tool returns a blocking issue, and Claude fixes it.
5. Claude shows you both captions and asks when to schedule. You reply "Tuesday 9 a.m. Eastern." Claude converts that to an ISO-8601 time with an offset and asks for your approval. Your Claude app may also show its own tool permission prompt. After you approve, Claude calls schedule_post.

The post now sits on the PostQuest calendar like any post you scheduled by hand. You can still edit or move it there.

How does approval work for posts Claude creates?

AI-created posts start as drafts. Claude needs your explicit approval to schedule a post, publish it immediately, or delete it. Publishing immediately also requires a workspace setting that is off by default. These rules come from three layers, and it helps to know which layer enforces what.

Layer	What it controls
PostQuest server rules	Workspace scope, platform validation before scheduling, publish-now blocked unless the workspace allows it, published posts cannot be deleted
PostQuest tool instructions	Tell Claude to create drafts by default and to ask before scheduling, publishing, or deleting
Claude's tool permissions	Claude asks before a tool runs unless you chose Allow always for it

The server-side rules hold no matter what Claude decides. The instruction layer depends on Claude following the server's guidance, which it does in normal use. The permission layer depends on your choices. Anthropic recommends clicking Allow always only for tools you trust to run unsupervised. I suggest keeping prompts on for schedule_post, publish_post, and delete_post.

To turn on immediate publishing, open PostQuest Settings → API & MCP and use Immediate agent publishing for the workspace. Scheduling works without it.

One caution comes from Anthropic's help article: in Claude's Research mode, Claude can call connector tools automatically without further approval. Switch off PostQuest's write tools before you start a Research session, or leave the connector off for that chat.

How secure is a Claude MCP connection?

Claude MCP access to PostQuest is limited by OAuth, by workspace, and by the published tool list.

• OAuth instead of passwords. You sign in on PostQuest's own page. Claude receives a token, not your PostQuest password, and never sees the credentials for your social networks.
• One workspace per authorization. On the consent screen you choose a single workspace. Claude cannot see posts or accounts in your other workspaces. Reconnect the connector to switch.
• Untrusted content stays data. The server tells Claude to treat post content and account names as data, not instructions. That reduces the risk that text inside a draft steers Claude into another action.
• Revoking access. Remove the connector in Claude under Customize → Connectors, or open PostQuest Settings → API & MCP and disconnect it under Connected OAuth apps. Disconnecting revokes the access and refresh tokens for that workspace grant. In Claude Code, use Clear authentication in /mcp.

What are the limitations of Claude MCP scheduling?

Claude MCP scheduling is useful, but it is not a replacement for every part of a social media tool.

• No AI generation tools in PostQuest's MCP server. Claude writes the captions itself. PostQuest's in-app AI assistant is a separate feature that needs a paid plan, and MCP calls do not use its AI credits.
• Media from your device needs a hop. Claude cannot hand a chat attachment to PostQuest. Use the upload widget, the upload link in Claude Code, or a public HTTPS URL.
• Plan limits still apply. The seven-day trial allows 20 scheduled or published posts, five of them on X. Paid plans start at $15 a month as of October 2026. Check current PostQuest plans before a large batch.
• Account connections happen in PostQuest. Claude cannot connect a new Instagram or TikTok account. Do that in the PostQuest app first.
• One workspace at a time. Agencies managing several clients need to reauthorize to switch workspaces.
• No best-time suggestions. PostQuest does not recommend posting times. Claude can read your post analytics and suggest patterns, but tools such as Buffer and Sprout Social have built-in best-time features.

If you work mostly in ChatGPT, the ChatGPT MCP setup connects to the same endpoint. Full MCP write actions in ChatGPT are limited to Business and Enterprise/Edu workspaces, according to OpenAI's developer mode help article (checked October 7, 2026).

How do you troubleshoot a Claude MCP connection?

Most problems come from sign-in, workspace state, or a request Claude formed incorrectly. The table covers the common cases.

Symptom	Likely cause	Fix
Claude never shows the PostQuest tools	The connector is off for this chat	Click + → Connectors and switch PostQuest on
Sign-in lands on email verification or onboarding, not the consent screen	The PostQuest account is not verified or onboarded yet	Finish both in PostQuest, then click Connect again
"This workspace trial or subscription has ended"	The trial expired or the plan lapsed	Choose a plan in PostQuest, then retry the request
You cannot add a custom connector on Free	Free plans allow one custom connector	Remove the unused connector or upgrade Claude
Team member sees no "Add custom connector" option	Only owners can add connectors on Team and Enterprise	Ask an owner to add PostQuest in Organization settings → Connectors
Scheduling fails with a time error	The timestamp has no time zone offset	Ask Claude to use ISO-8601 with an offset, or name the city
publish_post is refused	Immediate agent publishing is off for the workspace	Schedule the post instead, or enable the setting in Settings → API & MCP
A post fails validation	Caption length or media does not meet a network's rules	Ask Claude what validate_post returned, and check the platform specs
An account shows as needing reconnection	The network token expired	Reconnect the account in PostQuest; Claude cannot do this
Requests start failing during a large batch	Rate limit reached (60 requests a minute, 1,000 an hour per connection)	Wait for the retry time Claude reports, then continue

If none of these match, remove the connector and add it again. Anthropic's help article notes that editing a custom connector requires removing and re-adding it.

How do you make Claude part of a weekly routine?

Claude MCP pays off most when it supports a fixed rhythm instead of one-off requests. A simple pattern works: once a week, ask Claude to review last week's analytics, propose drafts for each content pillar, and validate them. Then you approve the schedule in one pass. My guide to building a consistent social media posting schedule covers how to choose the slots.

For network-specific details, read how to schedule Instagram posts and how to schedule LinkedIn posts. Both explain format rules that Claude will run into when it validates drafts.

When you are ready, open the Claude MCP setup page, copy the endpoint, and start with a draft-only prompt. You can review everything Claude creates before any of it goes live.